fosslight-binary 4.1.29__tar.gz → 4.1.30__tar.gz

{fosslight_binary-4.1.29 → fosslight_binary-4.1.30}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: fosslight_binary
-Version: 4.1.29
+Version: 4.1.30
 Summary: FOSSLight Binary Scanner
 Home-page: https://github.com/fosslight/fosslight_binary_scanner
 Author: LG Electronics

{fosslight_binary-4.1.29 → fosslight_binary-4.1.30}/requirements.txt

@@ -8,5 +8,5 @@ py-tlsh
 pytz
 XlsxWriter
 PyYAML
-fosslight_util>=1.4.41
+fosslight_util>=1.4.43
 dependency-check

{fosslight_binary-4.1.29 → fosslight_binary-4.1.30}/setup.py

@@ -33,7 +33,7 @@ if __name__ == "__main__":
 
     setup(
         name=_PACKAEG_NAME,
-        version='4.1.29',
+        version='4.1.30',
         package_dir={"": "src"},
         packages=find_packages(where='src'),
         description='FOSSLight Binary Scanner',

{fosslight_binary-4.1.29 → fosslight_binary-4.1.30}/src/fosslight_binary/_help.py

@@ -15,6 +15,7 @@ _HELP_MESSAGE_BINARY = """
         -h\t\t\t\t    Print help message
         -v\t\t\t\t    Print FOSSLight Binary Scanner version
         -s\t\t\t\t    Extract only the binary list in simple mode
+        -e <path>\t\t\t    Path to exclude from analysis (files and directories)
         -o <output_path>\t\t    Output path
         \t\t\t\t    (If you want to generate the specific file name, add the output path with file name.)
         -f <format>\t\t\t    Output file format (excel, csv, opossum, yaml)

{fosslight_binary-4.1.29 → fosslight_binary-4.1.30}/src/fosslight_binary/_jar_analysis.py

@@ -161,7 +161,7 @@ def get_oss_info_from_pkg_info(pkg_info):
     return oss_name, oss_version
 
 
-def analyze_jar_file(path_to_find_bin):
+def analyze_jar_file(path_to_find_bin, path_to_exclude):
     remove_owasp_item = []
     owasp_items = {}
     remove_vulnerability_items = []
@@ -212,6 +212,11 @@ def analyze_jar_file(path_to_find_bin):
             # Even if the oss info is from pom.xml in jar file, the file name will be .jar file.
             # But the oss info from pom.xml could be different from .jar file.
             bin_with_path = val.get("filePath")
+
+            if any(os.path.commonpath([bin_with_path, exclude_path]) == exclude_path
+                   for exclude_path in path_to_exclude):
+                continue
+
             if not bin_with_path.endswith('.jar'):
                 bin_with_path = bin_with_path.split('.jar')[0] + '.jar'
 

{fosslight_binary-4.1.29 → fosslight_binary-4.1.30}/src/fosslight_binary/binary_analysis.py

@@ -13,7 +13,6 @@ import yaml
 import stat
 from fosslight_util.set_log import init_log
 import fosslight_util.constant as constant
-from fosslight_util.write_txt import write_txt_file
 from fosslight_util.output_format import check_output_format, write_output_file
 from ._binary_dao import get_oss_info_from_db
 from ._binary import BinaryItem
@@ -24,8 +23,8 @@ from fosslight_util.cover import CoverItem
 _PKG_NAME = "fosslight_binary"
 logger = logging.getLogger(constant.LOGGER_NAME)
 
-_REMOVE_FILE_EXTENSION = ['png', 'gif', 'jpg', 'bmp', 'jpeg', 'qm', 'xlsx', 'pdf', 'ico', 'pptx', 'jfif', 'docx',
-                          'doc', 'whl', 'xls', 'xlsm', 'ppt', 'mp4', 'pyc', 'plist', 'dat', 'json', 'js']
+_REMOVE_FILE_EXTENSION = ['qm', 'xlsx', 'pdf', 'pptx', 'jfif', 'docx', 'doc', 'whl',
+                          'xls', 'xlsm', 'ppt', 'mp4', 'pyc', 'plist', 'dat', 'json', 'js']
 _REMOVE_FILE_COMMAND_RESULT = [
     'data', 'timezone data', 'apple binary property list']
 INCLUDE_FILE_COMMAND_RESULT = ['current ar archive']
@@ -41,13 +40,13 @@ _start_time = ""
 windows = False
 BYTES = 2048
 
-JAR_VUL_HEADER = {'BIN_FL_Binary': ['ID', 'Source Name or Path', 'OSS Name',
+BIN_EXT_HEADER = {'BIN_FL_Binary': ['ID', 'Binary Path', 'OSS Name',
                                     'OSS Version', 'License', 'Download Location',
                                     'Homepage', 'Copyright Text', 'Exclude',
                                     'Comment', 'Vulnerability Link', 'TLSH', 'SHA1']}
 
 
-def init(path_to_find_bin, output_file_name, format):
+def init(path_to_find_bin, output_file_name, format, path_to_exclude=[]):
     global _root_path, logger, _start_time
 
     _json_ext = ".json"
@@ -69,37 +68,41 @@ def init(path_to_find_bin, output_file_name, format):
 
         if output_file != "":
             result_report = output_file
-            bin_txt_file = f"{output_file}.txt"
         else:
             if output_extension == _json_ext:
                 result_report = f"fosslight_opossum_bin_{_start_time}"
             else:
                 result_report = f"fosslight_report_bin_{_start_time}"
-            bin_txt_file = f"fosslight_binary_bin_{_start_time}.txt"
 
         result_report = os.path.join(output_path, result_report)
-        binary_txt_file = os.path.join(output_path, bin_txt_file)
     else:
         logger.error(f"Format error - {msg}")
         sys.exit(1)
 
     log_file = os.path.join(output_path, f"fosslight_log_bin_{_start_time}.txt")
-    logger, _result_log = init_log(log_file, True, logging.INFO, logging.DEBUG, _PKG_NAME, path_to_find_bin)
+    logger, _result_log = init_log(log_file, True, logging.INFO, logging.DEBUG, _PKG_NAME, path_to_find_bin, path_to_exclude)
 
     if not success:
         error_occured(error_msg=msg,
                       result_log=_result_log,
                       exit=True)
-    return _result_log, result_report, binary_txt_file, output_extension
+    return _result_log, result_report, output_extension
 
 
-def get_file_list(path_to_find):
+def get_file_list(path_to_find, abs_path_to_exclude):
     bin_list = []
     file_cnt = 0
     found_jar = False
 
     for root, dirs, files in os.walk(path_to_find):
+        if os.path.abspath(root) in abs_path_to_exclude:
+            continue
         for file in files:
+            file_path = os.path.join(root, file)
+            file_abs_path = os.path.abspath(file_path)
+            if any(os.path.commonpath([file_abs_path, exclude_path]) == exclude_path
+                    for exclude_path in abs_path_to_exclude):
+                continue
             file_lower_case = file.lower()
             extension = file_lower_case.split(".")[-1]
 
@@ -128,21 +131,22 @@ def get_file_list(path_to_find):
 
 
 def find_binaries(path_to_find_bin, output_dir, format, dburl="", simple_mode=False,
-                  correct_mode=True, correct_filepath=""):
+                  correct_mode=True, correct_filepath="", path_to_exclude=[]):
 
-    _result_log, result_report, binary_txt_file, output_extension = init(
-        path_to_find_bin, output_dir, format)
+    _result_log, result_report, output_extension = init(
+        path_to_find_bin, output_dir, format, path_to_exclude)
 
     total_bin_cnt = 0
     total_file_cnt = 0
     db_loaded_cnt = 0
     success_to_write = False
     writing_msg = ""
-    extended_header = {}
     hide_header = {'TLSH', "SHA1"}
     content_list = []
     result_file = ""
     bin_list = []
+    base_dir_name = os.path.basename(path_to_find_bin)
+    abs_path_to_exclude = [os.path.abspath(os.path.join(base_dir_name, path)) for path in path_to_exclude if path.strip() != ""]
 
     if not os.path.isdir(path_to_find_bin):
         error_occured(error_msg=f"Can't find the directory : {path_to_find_bin}",
@@ -151,7 +155,7 @@ def find_binaries(path_to_find_bin, output_dir, format, dburl="", simple_mode=Fa
     if not correct_filepath:
         correct_filepath = path_to_find_bin
     try:
-        total_file_cnt, file_list, found_jar = get_file_list(path_to_find_bin)
+        total_file_cnt, file_list, found_jar = get_file_list(path_to_find_bin, abs_path_to_exclude)
         return_list = list(return_bin_only(file_list))
     except Exception as ex:
         error_occured(error_msg=f"Failed to check whether it is binary or not : {ex}",
@@ -163,31 +167,21 @@ def find_binaries(path_to_find_bin, output_dir, format, dburl="", simple_mode=Fa
     else:
         cover = CoverItem(tool_name=_PKG_NAME,
                           start_time=_start_time,
-                          input_path=path_to_find_bin)
+                          input_path=path_to_find_bin,
+                          exclude_path=path_to_exclude)
         try:
             # Run OWASP Dependency-check
             if found_jar:
                 logger.info("Run OWASP Dependency-check to analyze .jar file")
-                owasp_items, vulnerability_items, success = analyze_jar_file(path_to_find_bin)
+                owasp_items, vulnerability_items, success = analyze_jar_file(path_to_find_bin, abs_path_to_exclude)
                 if success:
                     return_list = merge_binary_list(owasp_items, vulnerability_items, return_list)
-                    extended_header = JAR_VUL_HEADER
                 else:
                     logger.warning("Could not find OSS information for some jar files.")
 
             return_list, db_loaded_cnt = get_oss_info_from_db(return_list, dburl)
             return_list = sorted(return_list, key=lambda row: (row.bin_name))
 
-            if return_list:
-                str_files = (x.get_print_binary_only() for x in return_list)
-                success, error = write_txt_file(binary_txt_file,
-                                                "Binary\tsha1sum\ttlsh\n" + '\n'.join(str_files))
-
-                if success:
-                    _result_log["FOSSLight binary.txt"] = binary_txt_file
-                else:
-                    error_occured(error_msg=error, exit=False)
-
             sheet_list = {}
             for item in return_list:
                 content_list.extend(item.get_oss_report())
@@ -201,10 +195,10 @@ def find_binaries(path_to_find_bin, output_dir, format, dburl="", simple_mode=Fa
                     logger.info("Success to correct with yaml.")
             cover.comment = f"Total number of binaries: {total_bin_cnt} "
             if total_bin_cnt == 0:
-                cover.comment += "(No binaries detected.) "
+                cover.comment += "(No binary detected.) "
             cover.comment += f"/ Total number of files: {total_file_cnt}"
             success_to_write, writing_msg, result_file = write_output_file(result_report, output_extension, sheet_list,
-                                                                           extended_header, hide_header, cover)
+                                                                           BIN_EXT_HEADER, hide_header, cover)
         except Exception as ex:
             error_occured(error_msg=str(ex), exit=False)
 

{fosslight_binary-4.1.29 → fosslight_binary-4.1.30}/src/fosslight_binary/cli.py

@@ -16,6 +16,7 @@ _PKG_NAME = "fosslight_binary"
 def main():
     global windows
     path_to_find_bin = ""
+    path_to_exclude = []
     output_dir = ""
     format = ""
     db_url = ""
@@ -30,6 +31,7 @@ def main():
     parser.add_argument('-o', '--output', type=str, required=False)
     parser.add_argument('-d', '--dburl', type=str, default='', required=False)
     parser.add_argument('-f', '--format', type=str, required=False)
+    parser.add_argument('-e', '--exclude', nargs="*", required=False, default=[])
     parser.add_argument('--notice', action='store_true', required=False)
     parser.add_argument('--no_correction', action='store_true', required=False)
     parser.add_argument('--correct_fpath', nargs=1, type=str, required=False)
@@ -55,6 +57,9 @@ def main():
     else:
         path_to_find_bin = os.getcwd()
 
+    if args.exclude:  # -e option
+        path_to_exclude = args.exclude
+
     if args.output:  # -o option
         output_dir = args.output
 
@@ -88,7 +93,7 @@ def main():
     timer.setDaemon(True)
     timer.start()
 
-    find_binaries(path_to_find_bin, output_dir, format, db_url, simple_mode, correct_mode, correct_filepath)
+    find_binaries(path_to_find_bin, output_dir, format, db_url, simple_mode, correct_mode, correct_filepath, path_to_exclude)
 
 
 if __name__ == '__main__':

{fosslight_binary-4.1.29 → fosslight_binary-4.1.30}/src/fosslight_binary.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: fosslight-binary
-Version: 4.1.29
+Version: 4.1.30
 Summary: FOSSLight Binary Scanner
 Home-page: https://github.com/fosslight/fosslight_binary_scanner
 Author: LG Electronics

{fosslight_binary-4.1.29 → fosslight_binary-4.1.30}/src/fosslight_binary.egg-info/requires.txt

@@ -8,7 +8,7 @@ py-tlsh
 pytz
 XlsxWriter
 PyYAML
-fosslight_util>=1.4.41
+fosslight_util>=1.4.43
 dependency-check
 
 [:"darwin" in sys_platform]